Selecting Optimal Feature Set in High-Dimensional Data by Swarm Search

Table 8

Comparison of individual performances of different algorithm combinations.

	FS-PSO			FS-BAT			FS-WSA			FS-Cfs
	FS-PSO-PN	FS-PSO-DT	FS-PSO-NB	FS-BAT-PN	FS-BAT-DT	FS-BAT-NB	FS-WSA-PN	FS-WSA-DT	FS-WSA-NB	FS-Cfs-PN	FS-Cfs-DT	FS-Cfs-NB
Average error; (per classifier)	0.11096	0.1515	0.23894	0.08776	0.16954	0.38748	0.08798	0.09966	0.2323	0.46562	0.58604	0.62314
Average error; (per algo. group)	0.167133333			0.214926667			0.13998			0.558266667
St. dev. error; (per algo. group)	0.065406582			0.154928957			0.080164471			0.082350811
Time consumption; (per classifier)	14881.90607	760.10988	500.44817	5891.07444	645.36168	305.9484432	6671.47066	16908.49644	47372.0138	≈0	≈0	≈0
Av. time con.; (per algo. group)	5380.821372			2280.794854			23650.6603			≈0
St. dev. time con.; (per algo. group)	8229.204933			3131.196153			21171.35171			≈0
% Dimension; (per classifier)	75.23341747	75.33096362	25.70943097	64.57907591	75.53820549	30.29258042	55.47927538	57.73169913	45.58275197	16.94849599	16.9489599	16.94849599
Av. % Dimension; (per algo. group)	58.75793735			56.80328727			52.93124216			16.94849599
St. dev. % dim.; (per algo. group)	28.62088765			23.60378812			6.462861923			≈0